
Xinference Path Traversal Allows Unauthenticated File Deletion on Windows
Attackers can overwrite or delete JSON configuration files in Xinference 4.0.1 and earlier by injecting backslash sequences into model registration requests.

Attackers can overwrite or delete JSON configuration files in Xinference 4.0.1 and earlier by injecting backslash sequences into model registration requests.

The National Vulnerability Database rates CVE-2026-107436 as a high severity issue that allows object injection via untrusted data deserialization.

A critical flaw in phpMyFAQ lets attackers register passkeys for any user, bypassing passwords and TOTP to seize control of accounts.

Unauthenticated attackers can trigger arbitrary shortcode execution via the HTTP_REFERER header in affected versions of the User Registration & Membership plugin.

Unvalidated resume tokens in the Python SDK allow attackers to inject CLI flags, executing arbitrary commands outside security boundaries in versions before 0.2.121.

A high-severity flaw in libiec61850 allows remote attackers to crash servers via malformed dataset requests, with a fix available in version 1.6.2.

A bounds check error in the gomarkdown package allows attackers to crash services rendering untrusted input with very short link destinations.

Opening a crafted guitar tablature file in TuxGuitar versions before 2.1.0 triggers Java class loading that executes arbitrary code with application privileges.

Researchers found that the XWorm RAT control panel allows attackers to overwrite system files and run code without logging in, using data found in infected clients.

A critical vulnerability in J4Care's medical imaging archive allows attackers to delete files from storage while keeping database records intact, risking data loss.

Bank of America spends hundreds of millions on remediation after receiving an early release of Anthropic's Claude Mythos Preview tool.

Criminals sell a scanner that runs on compromised OpenAI, Anthropic and Google keys, forcing victims to pay for the compute costs of attacks on third parties.

A high-severity flaw in OpenNextJS Cloudflare lets attackers bypass middleware authorization by exploiting inconsistent path decoding in cached routes.

A heap-based buffer overflow in the XkbChangeKeycodeRange function of X.Org Server enables local privilege escalation to root access.

A flaw in the X.Org Server glamor module allows local attackers to gain root access via a heap-based buffer overflow, with a vendor patch now available.

An insecure direct object reference in ServiceNext SerTec 1.00 allows authenticated users to access any account, including unencrypted credentials.

A high-severity flaw in InvoiceShelf lets attackers use expired or mismatched tokens to view private invoices and estimates from other companies.

External attack surface scanning reveals hidden digital assets that internal audits miss, exposing the true scope of your organization's exposure to remote threats.

Misconfigured parsers and outdated libraries allow attackers to execute arbitrary code on your server without needing valid credentials.

Zero-click exploits compromise devices without user interaction, making traditional security warnings and training ineffective against silent data theft.

A high-severity vulnerability in the Vert.x PostgreSQL client allows attackers to crash applications via crafted message headers before authentication completes.

A flaw in Rancher allows users to create projects with duplicate names across clusters, granting unauthorized access to resources in separate environments.

Treating exceptions as permanent fixes creates hidden debt. Proper exception handling requires strict time limits, active monitoring, and compensating controls to keep risk within bounds.

A flaw in Eclipse Vert.x allows attackers to bypass size limits via compressed gRPC messages, risking memory exhaustion in affected server versions.